You Ask, I Answer: Do Keywords Still Matter in an AI SEO World?

Shane asks, “With the changes that Google has made to its algorithm, is there still any validity to having a list of keywords and tracking how they are ranking? Do keywords still have a place in the SEO process?”

When we decompose a topic, we end up with words and phrases. It’s literally impossible to write about a topic without using words and phrases; writing about coffee inevitably requires discussion about beans, brewing times, roasts, grinds, etc. Where the challenge is for marketers is in the topical adjacency of keywords – focusing on too narrow a list reduces your opportunities. The research part of SEO has drastically changed, but the publishing great, relevant content has not.

Can’t see anything? Watch it on YouTube here.

Listen to the audio here:

Download the MP3 audio here.

Machine-Generated Transcript

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for watching the video.

In today’s episode, Shane asks, with the changes that Google is made to its algorithm.

Is there still any validity to having a list of keywords and tracking how they are ranking? Do keyword still have a place in the SEO process? This is good question because a lot of what’s been written, including some pieces I’ve written myself, I’ve talked about the important topics, document level topic, sentence paragraphs, and larger chunks of text that Google now uses thanks to deep learning capabilities to measure the importance the relevance of a piece of content.

Now, here’s the thing about that.

When you decompose a topic, when you decompose a document, you end up with words and phrases, right? It’s literally impossible to write about the importance of coffee shops, without coffee, I guess, without talking about things like you know, beans brewing times, blends grinds, particular types of brews, coffee dishes like espressos and cappuccinos and lattes, and mocha chinos and all that stuff, right? All those words, words and phrases.

And so there is still some value in seeing how relevant a pages based on a series of words.

Now, where things are different is that you’re not going to try and get a page to rank for one word or phrase, right, you may have a focus keyword, but there’s got to be a bunch of keywords around that that should also be on that page that are relevant to that page.

This is a concept in natural language processing called adjacency.

What words and phrases are adjacent to the primary focus keyword.

So again, with the discussion about espresso for example, espresso has a particular type of being a particular type of grind, method, machinery.

And so a page that is about espresso should also have those terms featured on it in some degree, right? Because that’s what the topic is about.

And so that page may and probably should have multiple keywords that’s being tracked for but they’re all related around that central concept.

But idea.

The best SEO tools allow you to see how a page rank for multiple keywords right, and you can set it up to design design, your monitoring for multiple keywords, words and phrases have obviously you can’t make content without them.

Now, what’s changed in the SEO process is not the words in the phrases per se, but the research right the research part of SEO is what has drastically changed in this artificial intelligence world and AI driven SEO What’s changed about that is that you need much more sophisticated tools that do natural language, proper language processing, to get those adjacencies.

And you need subject matter experts to talk to in order to discover those hidden terms that may not be frequent on a page, but indicate your expertise they indicate your authority, your relevance, the fact that you know the topic deeply.

So keyword frequency tools could be one way of doing that just determining how many times does a word or phrase up here, our friends over at Talkwalker have a basic frequency counter in a lot of their software.

Some of the SEO tools are starting to get there.

But your best results are always going to be from more advanced natural language processing software.

So you can use some of the cloud services like Amazon and Google and IBM all have a natural language processing capabilities, digest large quantities of text, I use an open source library that’s built into Watson.

It’s an add on to the our language, the our programming language in order to do a lot of that, as well, because with Google and other cloud services, you do pay a nonzero amount of money, it’s small, it’s not a lot of money, but for the effort, if you’re gonna, because you’re gonna have to write code to connect to those API’s Anyway, you may as well just build it with something doesn’t cost you money every time you write.

So, the research part is what has changed.

How do you determine those keyword adjacencies.

And there’s a bunch of different methods.

There’s, as I mentioned, basic term frequency counts, which actually are not bad for determining just how often does a term or phrase show up within a large body of copy.

There is a measure called inverse document frequency, which is part of term frequency that you can use to determine how many times does a term uniquely appear in the document that gets rid of things like the word law that offers no predictive power about expertise across a bunch of documents, but it shows up a whole bunch.

And then there are more advanced methods like vectorization, which allow you to mathematically gives you GPS coordinates on page two terms, that CEO who lives near that term, if you will.

And then there’s even more advanced models on that.

But those are sort of the three major methods for measuring those keywords, and they’re pretty good.

They will get you a good percentage of the way, especially if you don’t have domain knowledge.

One of the things that’s can be tough for a lot of SEO and content practitioners is that you need to have at least some level domain expertise in order to be able to craft great copy That the days of googling for two minutes and copying and pasting a bunch of stuff off of other pages are long gone.

It now because of natural language models being able to understand things like grammar and syntax to some degree.

That technique, which used to work in the old days does not work anymore.

So you do need that domain level expertise.

If you’re wondering how to make use of this with any don’t have machine learning capabilities, you don’t have data scientists on staff, your best bet is to sit down with your subject matter experts.

Get out a little portable audio recorder, turn it on, and have a conversation with them.

Put together a list of questions about your keywords about your topic, right.

And for 15 minutes, 20 minutes.

Work with your subject matter expert, talk to them and just ask them questions about the topic as though you are googling it but you’re googling it from inside somebody’s head.

So what is The best way to make espresso or what should I look for an espresso beans, and we’ll get to talk about the roast versus the grind and a lot of detail.

And that those conversations are what you can then start mining for that true expertise.

That’s the easiest way to get started in the new keyword landscape.

The fact of the matter, though, is that at the end of the day, the core goal of Google’s algorithm changes has not changed, which is to help you find great relevant content.

Right? So as long as you’re creating longer form, great relevant content with subject matter expertise, you will deliver on what is that Google is looking for, in its its search algorithms.

Worried less about individual keywords worry much more about topical expertise.

So use those use those subject matter experts.

Really good question.

Look into some of those natural language processing tools that I mentioned there.

There are a lot of capabilities out there and most of them are relatively inexpensive.

I would shy away from people offering like individual point solutions, because they will tend to be expensive and they will tend to be things that if you invested some time and money in yourself in developing the capabilities yourself, you would have them and could use them in an unlimited fashion, rather than paying you 599 a month for you know, a tool that does one thing, so be on the lookout for that.

As always, please leave your follow up comments and questions below.

Subscribe to the YouTube channel on the newsletter, I’ll talk to you soon take care want help solving your company’s data analytics and digital marketing problems? This is Trust insights.ai today and let us know how we can help you


You might also enjoy:


Want to read more like this from Christopher Penn? Get updates here:

subscribe to my newsletter here


AI for Marketers Book
Get your copy of AI For Marketers

Analytics for Marketers Discussion Group
Join my Analytics for Marketers Slack Group!


Pin It on Pinterest

Shares
Share This